As poll close draws near, and Obama seems positioned to win in enough early-closing states to block McCain’s path to the White House before things get late, it’s worth looking at the obvious question: does calling an election before all the votes have been cast have an effect on states where polls are still open?
This is a hard question to answer. A quick search of the literature brings an article by Seymour Sudman, in the Autumn 1986 issue of The Public Opinion Quarterly. After critiquing a variety of studies that attempted to find a West Coast lag in 1980 and 1984, Sudman concludes:
As with so many other interesting real-world events, measuring the exact effect of one thing, the exit polls, and controlling for everything else becomes very tricky. Based on a consensus of the data, there is a possibility of a small decrease ranging from 1 to 5 percent in total vote in congressional districts where polls close significantly later than 8 PM EST in those elections where the exit polls suggest a clear winner when previously the race had been considered close. No more precise estimate is possible with the data available.